Enhanced Vegetation Index Products Research Articles

Spatial and Temporal Variations of Vegetation Phenology and Its Response to Land Surface Temperature in the Yangtze River Delta Urban Agglomeration

In the Yangtze River Delta urban agglomeration, which is the region with the highest urbanization intensity in China, the development of cities leads to changes in land surface temperature (LST), while vegetation phenology varies with LST. To investigate the spatial and temporal changes in vegetation phenology and its response to LST in the study area, this study reconstructed the time series of the enhanced vegetation index (EVI) based on the MODIS EVI product and extracted the vegetation phenology indicators in the study area from 2002 to 2020, including the start of the growing season (SOS), the end of the growing season (EOS), and the growing season length (GSL), and analyzed the temporal–spatial patterns of vegetation phenology and LST in the study area, as well as the correlation between them. The results show that (1) SOS was advanced, EOS was postponed, and GSL was extended in the study area from 2002 to 2020, and there were obvious differences in the vegetation phenology indicators under different land covers and cities; (2) LST was higher in the southeast than in the northwest of the study area from 2002 to 2020, with an increasing trend; and (3) there are differences in the response of vegetation phenology to LST across land covers and cities, and SOS responds differently to LST at different times of the year. EOS shows a significant postponement trend with the annual mean LST increase. Overall, we found differences in vegetation phenology and its response to LST under different land covers and cities, which is important for scholars to understand the response of vegetation phenology to urbanization.

Carbon dioxide fluxes over irrigated and rainfed alfalfa in the Southern Great Plains, USA

The annual dynamics of carbon dioxide (CO2) fluxes for irrigated and rainfed alfalfa (Medicago sativa L.) in the Southern Great Plains of the United States of America (USA) under different watering regimes are not yet fully understood. The main objective of this study was to examine the dynamics of eddy covariance (EC) measured CO2 fluxes in relation to various biophysical factors and hay harvests for irrigated and rainfed alfalfa in central Oklahoma, USA. The study also aimed to investigate the relationship between CO2 fluxes and satellite-derived enhanced vegetation index (EVI) at different spatiotemporal scales and to assess the temporal variability in CO2 fluxes and EVI to variable growing conditions and hay harvests. The cumulative hay yields were 7.15 t ha−1 (two harvests in 2019) in the rainfed field and ∼9 t ha−1 (4–5 harvests in 2020 and 2021) in the irrigated field. Having sufficient rainfall during April and May was crucial to achieve economically feasible yields of alfalfa during the first harvest in May. The availability of water strongly regulated the potential for regrowth and carbon uptake of alfalfa following harvesting. The alfalfa fields were near carbon neutral or a small carbon source from January to mid-March and carbon sink after the initiation of vegetative growth in mid-March. The alfalfa fields were strong carbon sinks (cumulative annual net ecosystem CO2 exchange, NEE, up to −578 g C m−2 in irrigated field) on an annual scale. When accounting for the loss of carbon due to the removal of hay from the fields, the carbon balance of the alfalfa fields varied from small carbon sinks to small carbon sources, depending on the amount of hay harvested annually and the growing conditions. In general, the temporal patterns of CO2 fluxes and EVI were similar in relation to growing conditions and hay harvests. However, some discrepancies and time lags were observed due to the coarse spatiotemporal resolution of the EVI products. Thus, it is essential to integrate two or more satellite products with different temporal and spatial resolutions to accurately monitor the frequent and varying sizes of hay harvests and vegetation regrowth after harvesting, and to simulate continuous time-series CO2 fluxes.

Land cover and management effects on ecosystem resistance to drought stress

Abstract. Drought events are projected to become more severe and frequent across many regions in the future, but their impacts will likely differ among ecosystems depending on their ability to maintain functioning during droughts, i.e., ecosystem resistance. Plant species have diverse strategies to cope with drought. As a result, divergent responses of different vegetation types for similar levels of drought severity have been observed. It remains unclear whether such divergence can be explained by different drought duration; co-occurring compounding effects, e.g., heat stress or memory effects; management practices; etc. Here, we provide a global synthesis of vegetation resistance to drought and heat using different proxies for vegetation condition, namely the vegetation optical depth (L-VOD) data from the ESA Soil Moisture and Ocean Salinity (SMOS) passive L-band microwave mission and enhanced vegetation index (EVI) and kernel normalized difference vegetation index (kNDVI) from NASA MODIS. Due to its longer wavelength, L-VOD has the advantage over more commonly used vegetation indices (such as kNDVI, EVI) in that it provides different information on vegetation structure and biomass and suffers from less saturation over dense forests. We apply a linear model accounting for drought and temperature effects to characterize ecosystem resistance by their sensitivity to drought duration and temperature anomalies. We analyze how ecosystem resistance varies with land cover across the globe and investigate the potential effects of forest management and crop irrigation. We compare estimates of ecosystem resistance to drought and heat as retrieved from L-VOD, kNDVI, and EVI products. We find that regions with higher forest fraction show stronger ecosystem resistance to extreme droughts than cropland for all three vegetation proxies. L-VOD indicates that primary forests tend to be more resistant to drought events than secondary forests when controlling for the differences in background climate, but this cannot be detected in EVI and kNDVI. The difference is possibly related to EVI and kNDVI saturation in dense forests. In tropical primary evergreen broadleaf forests, old-growth trees tend to be more resistant to drought than young trees from L-VOD and kNDVI. Irrigation increases the drought resistance of cropland substantially. Forest harvest decreases the drought resistance of forests. Our results suggest that ecosystem resistance can be better monitored using L-VOD in dense forests and highlight the role of forest cover, forest management, and irrigation in determining ecosystem resistance to droughts.

Improved global 250 m 8-day NDVI and EVI products from 2000–2021 using the LSTM model

Satellite vegetation index (VI) products, such as normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI), have been widely used. However, they are severely contaminated by clouds and other factors and provide false signals of the surface vegetation conditions. In this study, the new global seamless 250 m, eight-day NDVI and EVI products from 2000–2021 were developed from Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance data using a long short-term memory (LSTM) neural network method. High-quality globally representative time series VI samples were constructed to train the model using a combination of the Savitzky-Golay filter (SG), Global LAnd Surface Satellite (GLASS) leaf area index (LAI) fitting and upper envelope methods. To evaluate the proposed method and the 250 m VI products, the MODIS VI product (MOD13Q1) was used for the inter-comparisons using four widely used VI reconstruction methods. Assuming that the MODIS VI data of high quality represents the true values, the root mean square error (RMSE) for NDVI and EVI generated by the LSTM model are 0.0734 and 0.0509, respectively.

Phenology estimation of subtropical bamboo forests based on assimilated MODIS LAI time series data

Phenology plays an important role in revealing the spatiotemporal evolution of forest ecosystem carbon cycles. The accuracy of vegetation phenology estimates based on remote sensing has improved in temperate zones. However, subtropical vegetation is complex, and the corresponding phenology estimates using remote sensing face great challenges. Bamboo forests are subtropical unique forest types and exhibit on– and off-years, fast growth, high productivity and carbon sequestration capability. In this study, we propose a new method to improve phenology estimates of bamboo forests by coupling the particle filter (PF) assimilation algorithm and a logistic model. The phenological metrics are estimated using high-precision leaf area index (LAI) assimilation products and a logistic model from 2001 to 2018, and the results are compared to those extracted from Moderate-Resolution Imaging Spectroradiometer (MODIS) LAI and the enhanced vegetation index (EVI) calculated based on the MODIS reflectance data. The results reveal that the R2 values between the start of the growing season (SOS) and end of the growing season (EOS) estimated by the assimilated LAI and ground-observed values are the highest (>0.50) and the root mean square errors (RMSEs) are the smallest (<6.35 days). A negative correlation occurs between the EVI-simulated and ground-observed SOS and EOS values, which indicates that EVI products cannot be adopted to estimate the phenology of bamboo forests. Compared to the MODIS LAI, the R2 values of the predicted SOS and EOS by the assimilated LAI data are improved by 3.67 times and 12.50%, respectively, and the RMSEs are reduced by 58.91% and 41.13%, respectively. Therefore, the new method solves the problem whereby the phenology of subtropical bamboo forests cannot be accurately extracted from MODIS LAI and EVI products. The temporal and spatial patterns of the SOS and EOS of bamboo forests are estimated with the new method from 2001 to 2018, and the SOS exhibits obvious spatial heterogeneity during on– and off-years, and the SOS during the on-years occurs slightly earlier than that during the off-years. A total of 70.13% of all pixels exhibit a SOS advance trend, while more than half of the areas (58.42%) present an EOS delay trend. The results indicate that coupling the data assimilation algorithm and phenology method greatly improves the estimation precision and reduces the estimation errors of the SOS and EOS of bamboo forests.

Assessing seasonal and long-term mangrove canopy variations in Sinaloa, northwest Mexico, based on time series of enhanced vegetation index (EVI) data

The use of remote sensing tools for mangrove monitoring is currently a strategy used to preserve this important coastal vegetation type. However, the temporal resolution of most satellite imagery limits the analysis of seasonal changes. Here, seasonal trend analysis (STA) was performed using an enhanced vegetation index (EVI) monthly time series (2005–2016) edited from MOD13Q1 products (MODIS Terra), to detect changes in mangrove cover in Sinaloa State in northwestern Mexico. The results were compared with a baseline map derived from a post-classification comparison of two mangrove distribution maps of 2005 and 2015. The STA procedure enabled the identification of mangrove canopy seasonal cycles, with open canopies between May and July and closed canopies from August to October. It was also possible to detect annual and long-term changes in canopy, with positive trends observed in most of Sinaloa’s mangrove ecosystems, which partially agree with the post-classification comparison results. Despite their coarse resolution, using the MODIS EVI products proved to be useful to confidently detect short- and long-term changes in mangrove cover, particularly in large ecosystems. Consequently, land use change analyses using these inputs could help to take actions related to mangrove cover management, considering the trends of change in this vegetation type.

A robust method for reconstructing global MODIS EVI time series on the Google Earth Engine

Obtaining high quality remote sensing vegetation index that is not noticeably affected by abiotic factors is critical for agricultural, ecological, climate and hydrological studies. Here we developed a computationally efficient and well-performed denoising method for reconstructing remote sensing vegetation indices, namely wWHd (weighted Whittaker with dynamic parameter λ in spatial). The single parameter λ is automatically estimated for every pixel by a multiple linear regression. Weights updating and the inherit nature of Whittaker make Whittaker robust for contaminations. We applied the wWHd on the Google Earth Engine (GEE) platform for reconstructing 500 m resolution enhanced vegetation index (EVI) time series from moderate-resolution imaging spectroradiometer (MODIS) at global scale and for the period of 2000–2017. To demonstrate its robustness, wWHd was compared with four well-known denoising methods, i.e. Fourier-based approach (Fourier), Savitzky-Golay filter (SG), Asymmetric Gaussian (AG) and double logistic (DL) at 16,000 randomly selected sites. All approaches were evaluated using two indices at each site: (1) root mean square error (RMSE) between observed best quality EVI and gap-filled EVI series, and (2) roughness of gap-filled EVI series. Results show that wWHd has an RMSE (indicating fidelity) of ∼0.032, which is similar to Fourier and SG at ∼90% sampled sites, but outperforms (∼0.02 less in the RMSE) AG and DL at ∼45% and ∼25% sampled sites. Among the four, wWHd has the lowest (best) roughness of ∼0.003. These performances demonstrate that wWHd balances fidelity and roughness well. Another advantage is that the wWHd is computationally more efficient than others, and is currently the only one denoising method deployed on the GEE. Our results suggest that it is promising to use the proposed wWHd method for processing remote sensing vegetation indices with high spatiotemporal resolution and the reconstructed EVI product should be widely used by global community.

An Improved Approach Considering Intraclass Variability for Mapping Winter Wheat Using Multitemporal MODIS EVI Images

Winter wheat is one of the major cereal crops in the world. Monitoring and mapping its spatial distribution has significant implications for agriculture management, water resources utilization, and food security. Generally, winter wheat has distinguished phenological stages during the growing season, which form a unique EVI (Enhanced Vegetation Index) time series curve and differ considerably from other crop types and natural vegetation. Since early 2000, the MODIS EVI product has become the primary dataset for satellite-based crop monitoring at large scales due to its high temporal resolution, huge observation scope, and timely availability. However, the intraclass variability of winter wheat caused by field conditions and agricultural practices might lower the mapping accuracy, which has received little attention in previous studies. Here, we present a winter wheat mapping approach that integrates the variables derived from the MODIS EVI time series taking into account intraclass variability. We applied this approach to two winter wheat concentration areas, the state of Kansas in the U.S. and the North China Plain region (NCP). The results were evaluated against crop-specific maps or statistical data at the state/regional level, county level, and site level. Compared with statistical data, the accuracies in Kansas and the NCP were 95.1% and 92.9% at the state/regional level with R2 (Coefficient of Determination) values of 0.96 and 0.71 at the county level, respectively. Overall accuracies in confusion matrix were evaluated by validation samples in both Kansas (90.3%) and the NCP (85.0%) at the site level. Comparisons with methods without considering intraclass variability demonstrated that winter wheat mapping accuracies were improved by 17% in Kansas and 15% in the NCP using the improved approach. Further analysis indicated that our approach performed better in areas with lower landscape fragmentation, which may partly explain the relatively higher accuracy of winter wheat mapping in Kansas. This study provides a new perspective for generating multiple subclasses as training inputs to decrease the intraclass differences for crop type detection based on the MODIS EVI time series. This approach provides a flexible framework with few variables and fewer training samples that could facilitate its application to multiple-crop-type mapping at large scales.

Integration of eddy covariance and process-based model for the intra-annual variability of carbon fluxes in an Indian tropical forest

Forests play a crucial role by regulating the global and local weather through the exchange of atmospheric gases and water vapor. The present study aims to study the intra-annual variability of net ecosystem exchange (NEE) of carbon dioxide in a sal (Shorea robusta) dominated moist deciduous forest in India by integrating eddy covariance (EC) data and Biome-Biogeochemical Cycle (Biome-BGC) model. The study also attempts to address the spatial variability of NEE with respect to phenology. Monthly average NEE were estimated using a calibrated Biome-BGC model and the spatial NEE was mapped using random forest (RF) regression algorithm. Phenology metrics were generated using the moderate resolution imaging spectrometer (MODIS) enhanced vegetation index product (MOD13A2) and its relationship with the estimated NEE was studied. RF regression model for monthly average spatial NEE estimation was built with an R2 of 0.84 and % RMSE of 3.68%. The study revealed that the NEE at the regional scale can be estimated using the basic meteorological variables like mean temperature, vapor pressure deficit, minimum temperature and total precipitation. Biome-BGC model output showed that the sal forest of the study area acted as a net sink of carbon in almost all months of 2015, except April to June. Peak NEE value (− 2.80 to − 2.96 g C m−2 day−1) was observed during October month. Annual NEE of sal forest in 2015 was found to be − 526.87 g C m−2 year−1. With the start of season (end of June), sal forest showed an increasing trend in NEE while decreasing trend was observed at the end of season (end of October). The study showed the applicability of Biome-BGC model in Indian forest when integrated with EC data. The study also highlighted the utility of RF in capturing the spatial variability of NEE over large area.

Assessing the ability of MODIS EVI to estimate terrestrial ecosystem gross primary production of multiple land cover types

Terrestrial ecosystem gross primary production (GPP) is the largest component in the global carbon cycle. The enhanced vegetation index (EVI) has been proven to be strongly correlated with annual GPP within several biomes. However, the annual GPP-EVI relationship and associated environmental regulations have not yet been comprehensively investigated across biomes at the global scale. Here we explored relationships between annual integrated EVI (iEVI) and annual GPP observed at 155 flux sites, where GPP was predicted with a log-log model: ln(GPP)=a×ln(iEVI)+b. iEVI was computed from MODIS monthly EVI products following removal of values affected by snow or cold temperature and without calculating growing season duration. Through categorisation of flux sites into 12 land cover types, the ability of iEVI to estimate GPP was considerably improved (R2 from 0.62 to 0.74, RMSE from 454.7 to 368.2gCm−2yr−1). The biome-specific GPP-iEVI formulae generally showed a consistent performance in comparison to a global benchmarking dataset (R2=0.79, RMSE=387.8gCm−2yr−1). Specifically, iEVI performed better in cropland regions with high productivity but poorer in forests. The ability of iEVI in estimating GPP was better in deciduous biomes (except deciduous broadleaf forest) than in evergreen due to the large seasonal signal in iEVI in deciduous biomes. Likewise, GPP estimated from iEVI was in a closer agreement to global benchmarks at mid and high-latitudes, where deciduous biomes are more common and cloud cover has a smaller effect on remote sensing retrievals. Across biomes, a significant and negative correlation (R2=0.37, p<0.05) was observed between the strength (R2) of GPP-iEVI relationships and mean annual maximum leaf area index (LAImax), and the relationship between the strength and mean annual precipitation followed a similar trend. LAImax also revealed a scaling effect on GPP-iEVI relationships. Our results suggest that iEVI provides a very simple but robust approach to estimate spatial patterns of global annual GPP whereas its effect is comparable to various light-use-efficiency and data-driven models. The impact of vegetation structure on accuracy and sensitivity of EVI in estimating spatial GPP provides valuable clues to improve EVI-based models.

Estimating mean air temperature using MODIS day and night land surface temperatures

Near surface air temperature (Ta) values measured by weather stations provide limited information about spatial patterns over a wide area. Remote sensing is a promising technology for providing a more accurate description of spatial variations in the Ta on both a regional and global scale. This paper presents a new approach for estimating mean Ta using a combination of MODIS day–night land surface temperatures (Ts) and enhanced vegetation index (EVI) data (called DTVX method). The advantages of the DTVX method include complete independence of ancillary data and non-contextual characteristics that circumvent the limitations inherent to the contextual temperature–vegetation index (TVX) method. Three land areas covered by the MODIS sinusoidal tile h26v04, h26v05, and h27v05 were selected as test areas because their terrain elevations vary in the range of several decameters to several kilometers. Based on the DTVX method, Terra MODIS 8-day day–night Ts and 16-day EVI products were used to obtain an 8-day average Ta from May 2010 to June 2011 (a total of 51 8-day periods). The daily average Ta values measured at 314 weather stations in the three tiles were utilized as the in-situ reference data. An RMSE value of 1.84 K with R 2 of 0.97 was observed for the Ta estimation in the h27v05 tile (plain area). In the case of the h26v05 tile (high mountain area), an RMSE of 2.45 K with R 2 of 0.96 was observed, whereas the h26v04 tile (mountain area) exhibited an RMSE of 2.34 K with an R 2 of 0.989. The RMSE of the three tiles was 2.23 K with an R 2 of 0.977. Sensitivity analysis indicates that the method developed herein is elastic with respect to the accuracy of day–night Ts and EVI data in areas with a greater EVI and day–night Ts difference, such as dense forests and high mountain areas.

Mapping forest damage in northern Nicaragua after Hurricane Felix (2007) using MODIS enhanced vegetation index data

This study employed a rapid assessment method for post-hurricane forest damage estimation following Hurricane Felix (4 September 2007) in Nicaragua, using Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m, enhanced vegetation index (EVI) imagery. Post-hurricane forest damage was characterized using pre- and post-hurricane EVI data, assisted by post-hurricane aerial photographs. The agreement between the obtained EVI damage map and the reference data set ranged from 91% in high-damage locations to 85% in low-damage locations, with 86% overall agreement. Broadleaf forests comprised more than half of the study area and experienced larger damage (>75% trees blown down) than Pine forests. Results indicate that MODIS 250 m EVI products are adequate for post-hurricane damage assessment of tropical forests.

Advanced Spaceborne Thermal Emission and Reflection Radometer (ASTER) Enhanced Vegetation Index (EVI) Products from Global Earth Observation (GEO) Grid: An Assessment Using Moderate Resolution Imaging Spectroradiometer (MODIS) for Synergistic Applications

We assessed the compatibility of three Advanced Spaceborne Thermal Emission and Reflection Radometer (ASTER) based Enhanced Vegetation Index (EVI) products generated in the GEO Grid system to Moderate Resolution Imaging Spectroradiometer (MODIS) EVI. The three products were two forms of the two-band EVI with ASTER red and NIR bands but without a blue band and the original, three-band EVI computed with ASTER red and NIR, and MODIS blue reflectances. Our assessment results showed good compatibilities of all the three ASTER EVI products with MODIS EVI, suggesting potential for synergistic applications of multi-resolution EVI.

Hurricane disturbance mapping using MODIS EVI data in the southeastern Yucatán, Mexico

This letter evaluated the use of Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m Enhanced Vegetation Index (EVI) standard product data (MOD/MYD13Q1 C5) to map the damage caused by Hurricane Dean (August 2007) to the forests in the Yucatán Peninsula of Mexico using a two-step vetting procedure. Sequences of MODIS EVI 16-day composite products captured before and after the hurricane were compared against 93 field damage plots to select an appropriate set of pre- and post-damage data. Aqua pairwise combinations produced the highest damage detection overall accuracy compared with Terra (82.4% vs. 73.8%, respectively) because of advantageous timing of the Aqua EVI compositing, relative to the hurricane event. The most accurate EVI damage map (91.4% overall) revealed highest damage detection in Saffir–Simpson hurricane wind scale zone 5 (i.e. wind speed >248 km h−1, i.e. 95%), followed by 93% in zone 4 (210–249 km h−1) and 87% in zone 3 (178–209 km h−1). Results indicate that MODIS EVI products provide timely and accurate maps of hurricane damage in subtropical forests.

Estimating root zone soil moisture at distant sites using MODIS NDVI and EVI in a semi-arid region of southwestern USA

This study investigates the potential of using Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) to estimate root zone soil moisture at native in-situ measured sites, and at distant sites under the same climatic setting. We obtained in-situ data from Soil Climate Analysis Network (SCAN) sites near the Texas-New Mexico border area, and NDVI and EVI products from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on board the Terra satellite. Results show that soil moisture values of the same depth are highly correlated ( r = 0.53 to 0.85) among sites as far as 150 km apart, and that NDVI and EVI are highly correlated at the same site ( r = 0.87 to 0.91). Correlation based on raw time series of NDVI and soil moisture is in general higher than that based on deseasonalized time series at every depth. The correlation reaches maximum value when vegetation index (VI) lags soil moisture by 5 to 10 days. NDVI shows a slightly higher correlation with soil moisture than EVI does by using the deseasonalized time series of NDVI and soil moisture. It is found that deseasonalized time series of NDVI and soil moisture are correlated at native sites ( r = 0.33 to 0.77), but not at sites where soil moisture is very low. Regression analysis was conducted using deseasonalized time series soil moisture and deseasonalized time series NDVI with a 5-day time lag. Regression models developed at one site and applied to a similar distant site can estimate soil moistures, accounting for 50–88% of the variation in observed soil moistures.

Wal-Mart from Space

We introduce an event data set of the location and opening dates for 3,043 Wal-Mart stores as a means for validating land-cover change-related products at medium (28.5 m) to coarse (250 m to 1 km) resolutions throughout the conterminous United States. Strengths of the Wal-Mart validation data set include: construction within most U.S. ecoregions, building footprints greater than a hectare in size, and construction dates that span much of the remote sensing record (1962 to 2004). We built the data set by geo-coding Wal-Mart addresses, establishing opening dates, and geolocating the footprints of 30 stores using online free highresolution (4 m) imagery. Disturbance events were evident at 25 Wal-Marts within a single scene of the Landsat Ecosystem Disturbance Adaptive Processing System (LEDAPS-beta) forest disturbance product. In addition, we found clear disturbance signals within two 16-day vegetation index time series: the 250 m Moderate Resolution Imaging Spectroradiometer normalized difference product (MOD44C) and the 1 km enhanced vegetation index product (MOD13A2).